Papers by Stephen D. Richardson
Neuron-Level Language Tag Injection Improves Zero-Shot Translation Performance (2025.acl-srw)
Copied to clipboard
| Challenge: | Language tagging is a method that trains models on specific language directions . injection is based on a language token embedded in the input layer . |
| Approach: | They propose a method whereby source and target inputs are prefixed with a unique language token and inject it into the input of every linear layer. |
| Outcome: | The proposed method improves translation performance with up to 2+ BLEU score point gain for certain language directions in a multilingual dataset. |
The Effects of Pretraining in Video-Guided Machine Translation (2024.lrec-main)
Copied to clipboard
| Challenge: | Existing approaches to improve VMT models integrate text and video modalities. |
| Approach: | They propose an approach that improves the performance of VMT models by using a new dataset which contains transcribed audio descriptions of movies. |
| Outcome: | The proposed model improves on the MAD (Movie Audio Descriptions) dataset. |